Beruflich Dokumente
Kultur Dokumente
#~B5
CIVIL ENGINEERING STUDIES
STRUCTURAL RESEARCH SERIES NO. 285
UNIVERSITY OF ILLINOIS
URBANA, ILLINOIS
NOVEMBER, 1964
Property of t!;9 reDsmme~t of
THEOHEriC:H i::;;; .~t;:;:_iLC ;'~iLGiiAriIGS
UNi~'I~~Sn ~ Df !LLifl~OiS
STUDIES OF THE
FATIGUE BEHAVIOR OF BUTT-WELDED JOINTS
IN HY-80 AND HY-100 STEELS
by
W. H. Munse
W. H. Bruckne r
A. J. Ha r't rna nn
J. B. Radzirninski
R. W. Hinton
and
J. L. Mooney
by
In Cooperation With
The Bureau of Ships, U. S. Navy
Contract NObs 88058
Project Serial No. SR-007-01-01, Task 856
UNIVERSITY OF ILLINOIS
NOVEMBER 1964
SYNOPSI S
in HY-80 steel containing internal defects are reported. The tests were con-
ducted on 3/4 in. and 1-1/2 in. thick material under stress cycles of zero-to-
tension and complete reversal. The types of weld defects studied include
porosity, slag inclusions, and artificially produced pores, and show that the
a given stress cycle when compared to sound welded specimens (with reinforcement
of fatigue cracks in welded joints. The results indicate that the first
more than half the test 1 ife. The subsequent crack propagation progresses
slowly and erratically at first, but the rate of growth is rapidly accelerated
from the time it reaches the specimen surface until complete fai lure occurs.
welded joints, and the behavior of these joints under conditions of fatigue.
restraint during welding are less critical than internal weld flaws (porosity,
steel was completed. Tests were carried out on plain plate specimens and trans-
verse butt welded joints in the as-welded condition using a zero-to-tension stress
cycle. Although the results for the plain plate specimens compared favorably
with similar tests on HY-80 steel, the fatigue behavior of the welded joints
in HY-100 was inferior to that of the HY-80 material. This is apparently due
TABLE OF CONTENTS
INTRODUCTION.
2.1 Material 3
2.2 Fabrication of Specimens 3
2.3 Description of Test Equipment. 4
2.4 Testing Procedure . . . 4
2.5 Radiographic Studies 4
VI I. SUMMARY 35
BIBLIOGRAPHY 39
TABLES . 40
FIGURES
LIST OF TABLES
Number
LIST OF FIGURES
Numbe r
Numbe r
Number
A.l RESULTS OF ULTRASONIC CRACK DETECTION STUDY FOR SPECIMENS TESTED
AT 50.0 KSI
A.2 RESULTS OF ULTRASONIC CRACK DETEG.TION STUDY FOR SPECIMENS TESTED
AT 0 TO +50.0 KSI
I. INTRODUCT ION
in HY-80 steel sometimes produced failures that initiated in the weld at.
study of the fatigue behavior of 3/4 in. thick defective welded joints in
HY-80 steel was undertaken in July 1962. In the present study, the effect of
internal weld flaws on the fatigue behavior of 1-1/2 in. thick HY-80 welded
joints has been evaluated and compared to that obtained for 3/4 in. specimens.
The second phase of the study has been concerned with a prel iminary
evaluation of the fatigue behavior of HY-IOO steel. Two types of specimens were
studied: as-rol led plain plate specimens, and butt welded joints in the
as-welded condition.
The studies reported herein were carried out on plain plate and
complete reversal. The report covers tests conducted during the period of
July 1963 to July 1964 and presents the resul ts of a number of large scale
fatigue tests.
Sixteen tests of 1-1/2 in. butt welded joints having slag and/or
number of tests of 3/4 in. butt welded joints were conducted to study the
internal weld flaw on the fatigue behavior of welded joints. These specimens
Fatigue tests were carried out on plain plate and butt welded speci-
mens fabricated from 3/4 in. HY-100 material. A zero-to-tension stress cycle
were used to study the effect of restraint on the development of internal weld
cracks and their subsequent effect on the fatigue behavior of butt welded
1.3 Acknowledgments
The program was carried out with funds provided by the Bureau of Ships, U.S.
Navy, under Contract 88058, Project Serial No. SR-007-01-01, Task 856.
of Metallurgical Engineering.
2.1 Ma te ria 1
this study; the HY-80 material was used in 3/4 in. and 1-1/2 in. thicknesses
while the HY-100 material was used in 3/4 in. thickness. The physical and
chemical properties of the materials are given in Tables 2.1 and 2.2, respec-
tive1y.
Mil 11018 grade. The electrodes were conditioned according to the requirements
give n i n Re f. ( 2) .
Figure 2.1 shows the geometry of the specimens tested in this test
program. The specimen blanks, 9 by 48 in., were flame cut from larger sections
of as-ro1 led HY-80 and HY-100 steel plates. The blanks used to fabricate
transverse butt welded joints were cut in half and the sawed edges were
angle. The welding procedures that were used are given in Figs. 2.2 and 2.3
(3/4 in. thick material) and Fig. 2.4 (1-1/2 in. thick material). All
welding was done in the flat position while the specimen blanks were clamped
passes.
were drilled and then the specimen blanks were machined to the desired
-4-
configuration (Fig. 2.1). In no case was material near the test section
removed by flame cutting. After being machined, the edges of the specimens
The tests were carried out using the University of III inois '
250,000 lb. lever-type fatigue machines (Fig. 2.5); one of the machines was
operated at 100 cpm while the other was operated at 160 cpm. There is a 15 to
force mul tipl ication provided between the dynamometer and the test specimen by
The throw of the eccentric determines the load range; i.e., the
difference between the maximum and minimum loads. The value of the maximum
load is set by means of the turnbuckle which is situated just below the
dynamome te r.
After the load had been set and the machine started, a microswitch
was set so that the machine would shut off when a crack had propagated ~rtially
through the specimen. Failure was assumed to have occurred when the micro-
switch shut the machine off. The microswitch was then disconnected and the
machine restarted; this was done to completely fracture the specimen so that
test i ng in order to determine the qual i ty of the transverse butt wel ds. All
of some 3/4 in. butt welded specimens of (HY-80) purposely provided with slag
Fatigue tests were carried out on plain plate and as-welded butt
joint specimens of HY-100 material. During the past few years, the use of
HY-100 has increased and as a result a knowledge of its fatigue behavior has
become important.
P100-1 l018-J, which is identical to procedure P80-1 1018-A used for the HY-80
The weldor and the welding procedure were qual ified in accordance with
requirements given in Ref. (5) for HY-80 material. The joint was able to
0
undergo a 180 guided side bend test without cracking. Figures 3.1 and 3.2
show typical micrographs for the HY-80 and HY-100 materials, respectively;
zero-to-tension stress cycle are tabulated in Table 3.1 and plotted in Fig. 3.3.
The S-N curve shown in this figure for HY-100 material was obtained using the
procedure out1 ined in Ref. (2). In addition, Fig. 3.3 contains the S-N curve
for 3/4 in. HY-80 plain p1ates(4). Comparing the curves it is noted that the
slope for the HY-100 material is greater than that for the HY-80 material.
near the radius of the test section rather than at the radius of the test
section. These failures initiated about 1/8 to 1/4 in. from the machined edge;
-7-
Fig. 3.4 shows this type of failure. The results tabulated in Table 3.1
indicate that the test 1 ives were not influenced by the mode of failure. One
of the specimens, HY-5, failed due to fretting in the upper pullhead after
396,500 cycles. The fretting cracks initiated at the drilled holes in the
pul lheads; examination of the area around the holes showed that the edges of
the holes were very rough. A new pullhead was welded to the specimen and a
Comparing the average S-N curves for the HY-lOO and HY-80 material,
one may note that although the fatigue strength of HY-lOO is greater than the
fatigue strength of HY-80 for lives of less than 150,000 cycles, the situation
zero-to-tension stress cycle are tabulated in Table 3.2 and plotted in Fig. 3.5.
Figure 3.5 also contains the S-N curve for 3/4 in. HY-80 as-welded butt
.. (4)
J 0 I nts . A1 though the 3/4 in. HY-80 spec i mens all failed due to the stress
concen t ra t ion of the ex te rna 1 wel d geomet ry, this was not true for the HY-lOO
the edge of the weld reinforcement because of very smal I defects which were
present in the root area of the weld. These defects (Fig. 3.6) were not visible
the fracture surface, it was difficult to locate the defects because of their
smallness.
It can be seen on Fig. 3.5 that there was a large amount of scatter
in the test results when failure occurred in the weld. In addi,tinn the failures
propagated rapidly when they originated in the weld. When failure occurred in
-8-
the weld, the fatigue 1 ives were less than those for HY-80 material by a
factor of about 8.
Even when failure was due to the external weld geometry, the fatigue
behavior of the HY-100 material was inferior to that of the HY-80 material.
Between 40,000 and 400,000 cycles the 1 ives of the two materials, at a given
Table 3.3 1 ists the fatigue strengths and slopes of the S-N curves
for 1-1/2 and 3/4 in. HY-80 and 3/4 in. HY-100 materials.
sl ightly greater than that of the HY-80 for short 1 ives (less than 100,000
cycles), it is sl ightly less than that of the HY-80 for longer 1 ives. In the
case of transverse butt welds, the behavior of the HY-100 joints is inferior
to that of the HY-80 joints and the slopes of the S-N curves for HY-100 material
are larger than the corresponding slopes of the HY-80 material, indicating
that the HY-100 steel is more notch sensitive in fatigue than HY-80.
the inc 1 us ion 0 f the we 1de d j 0 i n t s a s a fun c t ion 0 f 1 i fe for the HY- 8 and
HY-100 steels. The decrease in fatigue strength due to welding HY-100 is
3/4 in. HY-80 material, the reduction was 8 percent at N = 50,000 cycles and
The HY-100 butt welded joints were very sensitive to the presence of
internal defects. No tests were conducted with the weld reinforcement removed
since most welds would contain minute weld defec~s and 1 ittle or no increa.se in
specimens in two thicknesses of HY-80 steel (3/4 in. and 1-1/2 in.) using two
taken of each specimen and to remove any effect of the weld reinforcement on
The porosity and slag defects in the 3/4 in. thick specimens were
to produce the larger amounts of slag and porosity desired in the 1-1/2 in.
specimens. This was necessary because the increased thickness required larger
Specifications(5) .
were fabricated with artificial defects (steel balls) in the root area of the
The results of fatigue tests of 1-1/2 in. defective butt welds are
tabulated in Table 4.1 and plotted in Fig. 4.1. Since the tests were carried
out using two stress cycles, 0 to +70.0 ksi and 0 to +50.0 ksi, it was possible
to draw the S-N curve which is shown in Fig. 4.1. The slope, k = 0.21, is
larger than that for sound butt welds with weld reinforcement removed but is
less than that for butt welds tested in the as-welded condition. The S-N
curves for sound butt welds with and without weld reinforcement are also given
-10-
in Fig. 4.1 Comparing the three curves one may see that although the curve
for defective welds 1 ies entirely below that for welds with reinforcement
removed, it is below the curve for as-welded joints for 1 ives less than
150,000 cycles but above the curve for higher I ives. This indicates that at
long 1 ives the external geometry is more critical, with respect to fatigue,
than internal defects when the defects are of a size permitted by the code;
at the shorter 1 ives the internal defects produce greater effects than the
external geometry.
The geometry of the porosity in the 1-1/2 in. thick material was
similar to that of the 3/4 in. thick material which was discussed extensively
or ellipsoidal ~ Fig. 4.2a. The minimum distance from the surface of the weld
to the porosity ranged from 7/16 in. to 9/16 in. for the 1-1/2 in. thick
material and 7/32 in. to 5/16 in. for the 3/4 in. thick plate. All specimens
in the present test series with natural porosity were fabricated to produce a
single cluster of porosity. Specimens G-8l and G-88 each had fatigue cracks
that grew from the main cluster of porosity as weI I as from other small defects
located some distance from the cluster. Specimen G-8l had the lowest 1 ife
and specimen G-88 had the longest 1 ife of the specimens tested at 0 to +50.0 ksi.
The pore diameters in most of the test welds ranged from 1/64 in. to
-'-
3/64 in. The effective defect diameter", d , was determined for the largest
e
flaw in any cluster that propagated to failure and is tabulated in Table 4.2.
Figure 4.3 shows a plot of the percent of F 000 for a sound weld with the
IOO ~
reinforcement removed vs. the effective defect diameter; over the small range
Another parameter that has been used to specify the amount of porosity
is the percentage of defective area in the gross area of the specimen, A d '
re .
The defective area is defined as the area of porosity in the group or groups
s phe rica 1 manne r . The A d for a specimen was usually somewhat greater when
re .
based on the fracture surface than when based on radiographs because of the
The average A d for the 1-1/2 in. thick specimens was approximately
re .
equal to 0.16 percent; this is approximately the same as that obtained for
with increasing A . is evident for 0.02 < A . < 0.25 percent. This is
red red
unl ike the case for 3/4 in. material for which a trend of decreasing fatigue
Only two of the members failed to meet the Specification requirements but all
sound weld.
-12-
inclusions in the weld are tabulated in Table 4.3. Since the tests were
carried out using two stress cycles, it was again possible to obtain an S-N
curve for the test results. The test results are shown in Fig. 4.5. In
addition to the S-N curve for the defective welds, Fig. 4.5 contains the S-N
curves for sound transverse butt welds in the as-welded condition and with
the weld reinforcement removed. It is notable that the slope of the S-N curve
for welds with slag inclusions is the same as that for porosity specimens
Specifications(5). Unl ike the case for 3/4 in. thick material where an
there does not appear to be a similar trend for 1--1/2 in. material. As shown
between +37 ksi and +50 ksi, independent of the value of A d ' while the
re .
average fatigue strength for as-welded transverse butt welded joints is +46 ksi
and for sound welded joints with the reinforcement removed is +64 ksi. The
scatter for as-welded joints is +38 ksi to +48 ksi, which is essentially the
meeting the weld qual ity requirements of the Navy Specifications had higher
fatigue strengths than those which did meet the requirements (see specimens
G-91 vs. G-92 and G-93 in Table 4.3, and specimens Z-96 and Z-94 vs. Z-56, Z-74
the effect of internal geometry on the fatigue 1 ife becomes more important.
At this 1 ife, the upper bound of the results for defective welds is about 6 ksi
lower than the average fatigue strength of the sound as-welded joints.
artificial porosity (drilled holes filled with steel balls), were tested to
study the effect of the spacing and number of defects in aline on the fatigue
1 ife. However, it was not possible to obtain fully the desired information
since Z-lOO failed due to lack of fusion in the root area rather than at
drilled holes. The 1 ives for these three specimens are given in Table 5.1 of
Specimen Z-102, which had eleven 1/16 in. diameter bal Is spaced at
specimen with A d equal to 1.2 percent. This can be attributed to the widely
re .
distributed defect area and to the shape of the defects. Only six of the steel
balls were included in the initial fatigue portion of the crack which propagated
A number of 3/4 in. butt welded joints having internal weld defects
were tested using a 50.0 ksi stress cycle. The test results are tabulated in
Table 4.5; some of the complete reversal tests previously reported in Ref. (4)
are also included. The fatigue strength at 50,000 cycles has been plotted as
Furthermore, the upper bound of the scatter band for defective welds is about
12 ksi below the average fatigue strength for sound as-welded butt joints.
-14-
Fig. 4.10. The graph indicates that, for the range of intentional defects
defect and the amount of defective area. It is evident also that at 100,000
cycles the fatigue strength of the members with intentional flaws is about
the same as that of the as-welded joints with the reinforcement in place (no
intentional defects).
From the study reported above, it appears that the intentional internal
weld defects in specimens with the weld reinforcement removed are of more
importance than the external geometry of sound as-welded joints at shorter 1 ives,
less than 100,000 cycles, while the external weld geometry appears to be more
important at the longer 1 ives. It is also evident that the present code
requirements for porosity and slag will not guarantee a fatigue behavior for
1-1/2 in. thick butt joints comparable to that of a sound joint. Neither does
Unl ike the resul ts obtained for zero-to-tens ion tests of defective
3/4 in. butt welded joints, there does not appear to be any change in the
area, A
. re'd . ' for complete reversal tests .
-15-
The fatigue 1 ife of metals may be divided into two general categories:
_'~
given stress level, and the cycles of crack propagation from initiation to
failure. As part of the present study, nine 3/4 in. thick HY-80 transverse
butt welded specimens, Z-94 through Z-102 inclusive, were examined during
tested at 50.0 ksi while the remaining four were tested at 0 to +50.0 ksi.
Each of the specimens contained weld defects, the defects varying in type and
size as shown in Table 5.1. The data for three specimens from an earl ier
study(4) , Z-70, Z-79, and Z-80, are also included in the results.
III inois ' 250,000 lb. fatigue machine described earl ier. To study crack
initiation and propagation, two separate techniques were followed. The first
employed the use of the portable radiographic equipment shown in Figs. 5.1a
and 5. lb. The second method, used primarily for a qual itative comparison with
the radiographic technique, util ized the ul trasonic flaw detection equipment
size and location of internal weld defects. The specimen was next secured in
the fatigue machine and subjected to repeated loadings (the number of cycles
depending on stress cycle) at which time the machine was shut off with the
specimen being held at maximum tensile load. With the specimen in this
position, a radiograph was then taken of the test section. Fol lowing this,
0
the ul trasonic flaw detector, employing both a normal probe and a 45 angle
probe, was used to scan the specimen through the region in which the weld was
situated. For each of eight consecutive locations along the surface of the
test section, the ampl itude of the peak response as indicated on the detector
scope for a constant pulse energy was recorded. No attempt was made to
correlate these readings with actual flaw sizes; the method was used simply to
obtain prel iminary data concerning the feasibil ity of using such a device to
procedure above was then repeated. This process continued until such time as
a crack became evident, after which radiographs and ultrasonic readings were
graphic detection, of the initiation of fatigue gracks in the 3/4 in. thick
HY-80 butt welded specimens. The test specimens contained various types of
weld defects; these defects were generally placed in the second weld pass and
were located approximately midway between the specimen faces. The objectives
level, some direct correlation exists between the type or size of weld defect
and the initiation of a fatigue crack; and (2) to determine if there is some
-17-
simple relationship between the cycles to initiation and the specimen "test
;',
1 i fell. It should be clearly understood that, for the purposes of this study,
as shown by the radiographs using the procedure outl ined above. The significance
of correlating initiation with test 1 ife rather than cycles to failure will be
For the five specimens tested at 50.0 ksi, the 1 ife for fatigue
crack initiation varied from a low of 1,400 cycles for specimen Z-99 to a
maximum of 9,000 cycles for Z-94. A similar wide variation in cycles to init-
iation was exhibited by the specimens tested at 0 to +50.0 ksi. The 1 ife for
initiation ranged from 2,500 cycles for Z-lOO (an admittedly very poor specimen
containing large initial weld defect) to 106,000 cycles for Z-98. Data for
The large amount of scatter at both stress levels indicates that the
initiation of fatigue cracks. That such is the situation was evident from an
defect, al though not necessarily at the largest defect. This was in marked
contrast to 3/4 in. thick sound weld specimens (with reinforcement removed)
where fatigue cracks frequently had their origins outside the test section.
Test 1 ife - defined herein as the number of cycles necessary for a crack
to propagate to the surface of the test specimen.
-18-
The type and size of weld flaw leading to fatigue crack initiation
for each specimen is compared with cycles to initiation in Fig. 5.3. For
possible to determine the geometry at the extremity of the weld defects from
(1) size and geometry of weld defects; (2) number and type of defects in
weld area; (3) location of defects in specimen (i .e., proximity to other flaws
loading; and (5) the residual stresses in the weld metal. Without more data,
all factors using a statistical approach. The problem is further compl icated
ship between the cycles to initiation and the test life. At 50.0 ksi, the
ratio of cycles to initiation to test 1 ife varied from 0.264 for specimen
Z-99, to 0.833 for Z-94 (Table 5.1). For the specimens tested at 0 to
+50.0 ksi, the variation in this ratio ranged from 0.462 for specimen Z-80
initiated at more than half. the test 1 ife, with the average for all specimens
being 0.69. Thus, on the basis of these tests and the crack detection procedure
used, the number of cycles necessary to propagate the fatigue cracks to the
specimen surface was found to be less than 50 percent of the test 1 ife.,
defect in the 3/4 in. thick plate specimens can be divided into two reasonably
distinct regions. The first stage begins with the initiation of the crack
and continues until the crack approaches or actually intersects the specimen
surface (test 1 ife). It is during this period that the specimen experiences
true "fatigue crack" propagation, with the crack growing radially from the
mode of propagation in the joints changes to one in which the crack progresses
along planes obl ique to the loading direction, exhibiting the so cal led
The rate of propagation is rapidly accelerated during this latter stage and
ends with complete fracture within a relatively few cycles. This increased
rate of growth is due in large part to the higher stresses created as the
remaining net specimen section required to resist the load is decreased. The
Fig. 4.8.
crack in a plate specimen of the type used in this study, thus provid,ing a
-20-
consistent bound for evaluating the radiographic data regarding such crack
growth. It was the purpose of this phase of the project to investigate both
the fatigue and shear-type crack growth mechanisms and to cor/elate duration
for the specimens tested at 50.0 ksi, varied from 1000 cycles for specimen
Z-97, to a maximum of 4,500 cycles for Z-96. There was an apparent general
trend but no clearly defined consistent ratio between duration of fatigue crack
growth and specimen 1 ife. Rather, Fig. 5.4 shows that the number of cycles
of propagation falls within a fairly narrow band regardless of the total 1 ife
of the specimen. This observation is even more pronounced for the tests
performed at 0 to +50.0 ksi, when all of the tests are considered. However,
as shown in Fig. 5.5, the cycles of fatigue crack propagation for the current
series of tests ranged from 2,700 for specimen Z-lOO to 14,000 for Z-98, and
1 ife, although the increase is not great. The specimens from the previous
tests had greater durations of propagation even though their total 1 ives were
lower.
a fatigue crack forms and begins to extend radially outward from the defect,
the geometry at the tip of the crack becomes similar in most cases. The
up to the point of intersection with the specimen surface, when the mode
-21-
were located near the center of the weld in the specimen and all specimens in
the group tested were 3/4 in. thick, it appears reasonable that they should
mation presented in Figs. 5.4 and 5.5. The sl ight increase in duration of
propagation with test 1 ife may be due to such factors as variation in flaw
the actual rate of crack propagation, during both the normal fatigue and shear-
type growth stages. The results of measurements from the radiographs are
presented in Figs. 5.6 and 5.7 for tests performed at 50.0 ksi and 0 to
+50.0 ksi respectively. The data are referred to test life, thus different-
iating the two separate stages of propagation. Although the scatter band for
crack length prior to the test 1 ife appears to indicate a uniform rate of
behavior in the early stages of fatigue crack propagation. The limited number
in some specimens for a time beyond the appearance of the crack on the specimen
surface. Thereafter, the growth rate accelerated rapidly and failure usually
occurred within few additional cycles. Figures 5.6 and 5.7 do~ however, show
propagation at each stress cycle, two possibil ities are offered. First, the
planes that the crack fol lows after reaching the specimen surface are often
dictated by the location of other weld flaws in the near vicinity. The crack
will tend to grow toward these defects, especially if they, too, have
often passes near or into the specimen heat affected zone and fol lows that
1 ine to failure. Depending upon the crack path taken, the crack will progress
through the test section at a slower or more rapid rate. A second possibil ity
A crack situated toward the edge of the specimen would tend to induce bending
as well as axial stresses in the rema'ining net section, thus increasing the
wedging action at the tip of the crack and further stimulating the rate of
propagation.
5.5 Conclusions
been drawn regarding the initiation and propagation of fatigue cracks in 3/4
in. thick HY-80 transverse butt wel~ed specimens containing weld defects:
of a weld defect, the duration of its propagation to test 1 ife falls within a
-23-
fairly narrow range of cycles for a given stress level. This period defines
crack propagation ranged from 1,000 to 4,500 cycles for specimens tested at
50.0 ksi and from 7,000 to 14,000 cycles (excluding Z-lOO) for those tested
3. Beyond the point at which the crack reaches the specimen surface,
the increased stresses near the tip of the crack and throughout the entire
remaining net section cause a rapid increase in the rate of crack growth lead-
ing to failure within a relatively few cycles. During this period of crack
propagation,the path of the crack usually progresses along planes obl ique to
the direction of loading. This pattern then continues until complete fracture
behavior of butt welded joints in HY-80 steel fatigue crack propagation was
observed. These fatigue cracks propagated either along the fusion 1 ine~ in
the weld metal or in the base metal. Frequently the crack propagation in
the base metal was observed along metal-inclusion boundaries. In a few cases,
the inclusions were melted near the fusion 1 ine leaving small hot cracks in
that region.
of causing early fatigue failures when they produced hot cracks in the base
metal near the fusion 1 ine of the weld. These inclusions were tentatively
identified as sulfides because of their low mel ting point and their metal 10-
graphic appearance.
we 1dment.
the stresses during welding on the initiation of cracks in the weld deposit
the red uc t ion 0 r eli min a t ion 0 f hot c r a c kin gin the de po site d we 1d me tal .
Restraint was appl ied to the HY-80 steel during butt welding in
samples were cut in 3/4 in. and ]-1/2 in. thick HY-80 steel plates, with a
1/8 in. spacing between the plates at the root. Two 4~1/2 in. x 9 in.
plates were then placed in a 17 in. square block of 2 in. thick plain carbon
steel as shown in the diagram in Fig. 6.1. Both HY-80 steel plates were
then fil let welded into the plain carbon steel plate at both surfaces to
HY-80 plates only in the direction parallel to the long axis of the butt
weld. After the butt weld was completed, it was allowed to remain under
restraint for a minimum of 24 hours after the butt welding. The fi1 let
welds were initially checked by dye penetrant for possible surface cracking
etching. The length and position of the hot cracks were recorded. TL_
I flt!
sample was then repol ished and the hot cracks were again observed in order
plate is shown in Fig. 6.2. The test plates were cut into 9 in. wide strips
for test purposes. The butt weld was approximately 8 inches long in the 9 in.
"VII groove? but only the middle 6 inches of each butt weld was examined for
hot cracks. Metallographic sections were cut from the weld as shown in
Fig. 6.3 after each butt weld was held in restraint for a minimum time of
-26-
of 400 diameters,
The HY-80 steel was received in the quenched and tempered condition
analysis for manganese, phosphorus and sulfur in each plate near the 3/4 in.
the resul ts found in Table 6.1 will be used to represent the G-67, G-68, G-69,
and Z-87 plates, The G-69 plate was cut down from a 1-1/2 in. thick plate by
removing 3/8 in. from each side in order to obtain a 3/4 in. thick plate of
a higher sulfur composition. The Z-87 plate had been rol led down to its
The G plates reprepent a higher sulfur content and were taken from
the bottom center of the ingot (plate A-5) as shown in Fig. 6.2. Since the
width and thickness for the G-69 and Z-87 plates was made at 300x.
widths of the inclusions in the Z-87 plate were greater than those in the
G-69 plate on the average~ but the thickness of the inclusions in the Z-87
sample was less than those of the G-69 sample. If an average area of
-27-
a rectangular area, the average inclusion area in the Z-87 plate was 165 square
microns compared to 83 square microns for the G-69 plate inclusion area.
Table 6.2 1 ists the resul ts of this study and shows the difference in the
inclusions in the G-69 and Z-87 plates. Another approach taken to describe
measured within a selected area of the pol ished and etched sample. These
histograms of size distribution are shown in Figs 6.5 and 6.6. The histogram
in Fig. 6.5 shows two or more peaks in the width measurements because some
inclusions were discontinuous and they were measured separately. The only
difference observed between the width in the Z-87 and G-69 histograms is that
the inclusions are pinched off more frequently in the Z-87 samples.
inclusions. An inclusion was usually located near the fusion 1 ine where the
crack was formed. The cracks that are associated with inclusions do not
and thickness measurements in samples G-69 and Z-87 also indicate that the
geometric size or shape of the inclusions do not affect the crack frequency
6.5 Discussion
The hot cracks referred to in this report are the small cracks
radiating from the inclusions that have partially melted near the fusion
1 i neon the bas e me tal sid e . In some instances isolated flaws were observed
in the weld metal such as slag inclusions, porosity, cracking along fingers
of the weld metal and a few relatively large cracks not associated with
The hot cracks were associated with inclusions because they were
found either radiating from an inclusion near the fusion 1 ine or they were
in 1 ine with stringer laminations on the base metal side of the weld. Examples
of both these conditions are shown in the photomicrographs of Figs. 6.7 and
6.So By sectioning the welds in both the transverse and longitudinal directions
the plane of the hot cracks was observed. The longitudinal section was along
the long axis of the weld and parallel to the fusion 1 ine. The hot cracks were
found to extend less than 0.025 mm from the fusion 1 ine into the base metal
in most cases. Most hot cracks extended from 0.025 mm to D~D5 mm along the
long axis of the weld, The plane of the fractured surface of the hot crack
on 1-1/2 in, thick HY-SO steel samples and no hot cracking was observed in
me tal log rap h i c sec t ion s 0 f the s e we 1d s . I nan a t t em p t to c rea tea dd i t ion a 1
shrinkage stresses in the butt welds 1-1/2 in. x 1 in. x 1 in. blocks were
placed at the ends of the original 9 inch long butt weld double "V" groove in
order to prevent transverse shrinkage. few hot cracks were observed in this
procedure section was successful in causing numerous hot cracks in the G-69
samples. Partial restraint of the butt weld, resulting from only fillet
welding the HY-SO plates in a direction parallel to the long axis of the butt
weld, also caused hot cracks. The amount of cracking per transverse sample
In the above tests only the 1 -1/2 in. thick G-69 plates of higher
sulfur content (0.0155% S) were examined. The following tests were conducted
on 3/4 in. thick plates butt welded by the same procedure. One G-69 plate was
cut down from its 1-1/2 in. thickness to a 3/4 in. thick plate and welded to a
rolled 3/4 in. thick Z-S7 plate with a lower sulfur content (0.011% S). The
unrestrained 3/4 in. plate 1 isted in Table 6.3 was a 1-1/2 in. thick HY-SO
steel plate cut down from plate G-67 by machining 3/4 in, from only one side.
No restraint was used in welding the cut down G-67 plate and only a few hot
cracks were observed in the numerous samples prepared (Table 6.3). The
restrained, 3/4 in. thick sample in Table 6.3 was made from a 3/4 in. thick
plate cut down from a 1-1/2 in. G-69 plate by machining off 3/S in. of the rol led
surface from both sides. This plate was welded to a rol led down 3/4 in. thick
Z-S7 plate. The striking result of the butt welded G-69 and Z-S7 plate was
that almost all the hot cracking was on the high sulfur side of the fully
Tab 1e 6. 4 1 i s t s the ave rage hot c r a c k 1eng t h sob s e rve din the t ran s -
verse sections for all the welding conditions previously discussed for both the
1-1/2 in. and 3/4 in. thick samples. Most of these small hot cracks are about
0.025 mm long. Large cracks from other types of welding flaws are not included
in the values given in Tables 6.3 and 6.4, The relatively consistent hot crack
length can be related to the I imited depth to which inclusions melt beyond the
rare instances and an example of this type of cracking is shown in Fig. 6.9.
-30-
Cracks were also observed to initiate from the periphery of slag inclusions,
porosity (Fig. 6.10) and Ilfingers" of weld metal in the base metal (Fig, 6.11).
These cracks were usually larger than the hot cracks and Table 6.5 1 ists the
samples in which these defects were found along with their crack lengths,
f. Inclusion Study
the association of hot cracks with inclusions in the HY-80 steel microstructure.
Sulfides were assumed present from the results of a sulfur print which was
with an index of refraction between 1.35 and 1.63c This result was based on
Becke 1 ines observed around the edges of a large number of extracted stringers
particles matched the ASTM pattern for iron sil icate. The sulfides were probably
dissolved in the solvent used for extracting inclusions since they were not
Two sections of the 1-1/2 in. thick G-69 plate were joined under ful 1
one section was placed in the full restraint apparatus it was Ilbuttered 'i
(coated with Ell018 weJd metal by weaving the electrode over one side of the
The excess Ilbuttered il weld metal was machined off leaving a layer
0
30 scarf.)
of weld metal approximately 1/8 in. thick. The 1-1/2 in. thick sections were
then butt welded, sectioned and observed, The unbuttered side of the weld
had an average crack frequency and length similar to other 1-1/2 in. samples
welded in full restraint. The only cracks observed on the 'Ibuttered 'l side of
-31-
"the butt weld were near the root of the weld where i'buttered" weld metal had
not been placed. This clearly demonstrated that hot cracks can be decreased
to a low level by "buttering" the HY-80 steel before using regular welding
procedures provided the remaining buttered layer of weld metal is thick enough
"buttered" and normal G-69 plate shows the different Diamond Pyramid Hardness
(DPH) in the Heat Affected Zone (HAZ) of the two plates and al so in the
remaining "buttered " weld metal left after machining. The maximum hardness in
the HAZ of the "buttered" section was lower than maximum hardness in the
unbuttered HAZ of the HY-80 steel butt weld. A hardness survey of the HAZ of
the buttered scarf is shown in Fig. 6.13. This hardness survey shows the
effect of the large heat input caused by the extensive weaving motion of the
A G-69 plate, 1-1/2 in. thick, was machined down to four 3/4 in. thick
plates by removing 3/8 in. from both sides giving a thinner plate of higher
sulfur content (0.0155% S), These plates were butt welded in full restraint to
make two samples. The samples were then tested in a method similar to a
regular 3/4 in. thick fatigue test samplen These samples were tested in 0 to
+50.0 ksi tensile stress fatigue cycles. One sample reached a fatigue 1 ife of
170~000 cycles and the other 280,000 before an attachment weld at the end of
each sample broke. The fu11y restrained welds containing hot cracks were not
broken. The attachment weld was repaired for the sample that reached 280,000
cycles and the test was continued. The attachment weld broke twice before a
total of 430,000 cycles was reached. The G-69 sample welded in restraint had
probable number of hot cracks contained in the 4 in. wide 3/4 in. thick fatigue
sample used in the test just described. From measurements reported in Table 6.3
each transverse section of a cut down 3/4 in. G-69 sample welded in restraint
should contain one (1) hot crack on the average. If it is assumed that each
hot crack extended along the long axis of the weld for 20 mils. (0.020 in.)
and no hot cracks overlapped along the long axis of the weld, then the number
of hot cracks is equal to 4 in./0.020 in. = 200 cracks in a 4 in. wide, 3/4 in.
thick test sample. The 430~000 cycle fatigue 1 ife of the fully restrained weld
sample approaches the fatigue 1 ife of the unwe1ded plate. Thus, the hot cracks
along the fusion 1 ine do not appear to have an effect on zero=to=tension fatigue
Another 1-1/2 in. HY-80 steel sample was welded in restraint caused
by two 15 in. long plain carbon steel bars 3 in. square that were fillet welded
along the sides of the G-68 plate. The 1-1/2 in. thick plate was fillet welded
as if it were a 3/4 in. thick plate. The restraining bars were removed after
twenty-four hours and the entire plate was cut down to a 3/4 in. fatigue
sample. This sample was fatigue tested at a stress level of 50.0 ksi. It
failed after 13,000 cycles compared to a good weld fatigue 1 ife of about
60,000 cycles at the same stress level. The sample was sectioned and pol ished
across the fatigue fracture region. Fig. 6.14 shows the fractured region after
13,000 cycles. The arrows indicate the cut of the transverse section shown in
Fig. 6.15. The transverse section is near the center of the fatigue fracture.
The entire crack is in the weld metal in this region. Hot cracking in the HAZ
this center of the fatigue crack and centers of fatigue cracks in the weld metal
on the opposite side of the weld about 1/4 in. from the fracture, porosity was
observed. This pore was located near the root pass of the weld and very close
to the fusion 1 ine. This is a common location for small pores in the 3/4 in.
and 1-1/2 in. thick HY-80 steel butt welded plates. This pore was too small
removed. Two sma11 cracks were observed radiating out from each end of the pore
as shown in the metallographic section in Fig. 6.16. The cracks were nearly
0 0
parallel, within 4 , with one crack direction making an angle of 85 to the
0
fatigue stress di rection and with the other crack making an angle of 87.5 to
the stress direction. Further proof that these were fatigue cracks rather than
shrinkage cracks was the fact that at a high magnification the cracks were
6.6 Conclusions
in the HAZ of HY-80 weldments. The hot cracks are on the average 0.025 mm in
cracking. Hot cracks were observed for 0<0155% S, but practically no hot
cracks were observed for 0.011% S HY-80 steel weldments. However, the size
(width and thickness) of the inclusions does not appear to be related to the
hot crack frequency from the measurements of inclusion size made in this
investigation.
-34-
sl ightly lower maximum hardness in the HAZ of the "buttered" HY-80 steel
to reduce the fatigue 1 ife to any great extent for weldments tested in tension
from to+50.0 ksi for the 0.0155% S HY-80 steel welded in full restraint. The
long axis of the weld was perpendicular to the axis of tension and the hot
raiser factor.
cracking in the weld metal of the HY-80 steel sample G-68 subjected to 13,000
cycles of a reversal stress level of -50:0 ksi to +50.0 ksi. Porosity in the
weld metal was bel ieved to be responsible for the early fatigue failure of
VII. SUMMARY
reported here in. Ali mj ted number of tes ts were cond'Jcted and many of the
are a number of observations that are considered to be very signific~nt and are
Fatigue tests carried out on plain plate and butt welded specimens
of HY-l00 indicate:
1. The axial fatigue strength of ~he HY~100 steel plate was not
HY-IOO were found to have a lower fatigue resistance than comparable joints
From the fatigue tests carried out on butt welds with intentional
defective area.
flaws is about the same as that of the as-welded joints with the reinforcement
are more important than the external geometry of some as-welded joints, while
the external weld geometry appears to be more important at the longer 1 ives.
butt welded joints of HY-80 steel. Because of the lack of sensitivity and
weld defects, the location of the defect with respect to the cross section of
found to fall within a relatively narr~~ range of cycles for a given level of
the number of cycles necessary to propagate the fatigue crack to the specimen
the effects of inclusions and welding restraint upon the qual ity of weld metal
-37-
in butt welds of HY-80 steel e The results of these studies may be summarized
as follows:
several general observations may be made concerning the behavior of HY-80 and
HY-I00 steels and welded joints in these steels. These conclusions are as
fol lows:
shown in the fol10\'Jing tabulation plain plate members of !-IY-80 and HY ... I00
are found to have fatigue strengths considerably above that of mild structural
steel. However, the increase in fatigue strength for the butt welded joint
range between 60 and 87 percent for the HY-I00 and HY-80 materials respectively.
~"'ere wi thi n Navy spec i f i cat ions 1 imi ts for poros i ty and slag. HO\-1ever II when
the \oJe 1d re i nforcement was removed from members hav i ng fl aV'/s with j n the
than that of sound as-welded butt joints ",lith the reinforcement in place .
and detect ion in HY-80 suggest that th i s detect i on method if it can bl~ full y
developed, may provide earl ier crack detection than radiographic ex~mination.
HO\'Jever, this can only be real ized through extensive developments in the
operational techniques.
less effect on the fatigue resistance of t:le members than trie porosity and
BIBLIOGRAPHY
7. Lip sit t, H. A., For be s, F. W. and Ba i rd, R. B., II Crack Pro p a gat i on i n
Cold Rolled Aluminum Sheet," Proceedings, American Society for
Testing and Materials, 1959.
TABLE 2.1
Heat De sign at ion Thickness Yi e 1d ";', Tensile Elong. Reduction Charpy V-Notch
Numbe r (inches) Strength Strength in 2 inches in Area ft~lbs.
(ks i) (ksi) (pe rcent) (percent) @ -120 o F
I',
0.2 percent offset
-41-
TABLE 2.2
RESULTS OF FAT I GUE TESTS OF HY-l 00' PLA I N PLATE SPEC I MENS
(Zero-to-Tension)
...}(
Computed Fatigue Strength~ ksi
Spec i men Stress Life Locat ion
Numbe r Cycle (cycles) of ,1\,', F50 ,000 F100 ,000 F200 ,000 F500 ,000
(ksi) Fracture
if,
k=0.30l
,.,,',
a: failure initiated at r~dius of test section
f: failure initiated at mill scale surface near radius of test section
TABLE 3.2
II:;
Computed Fatigue Strength, ksi
Spec i men Stress Life Loca t ion
Numbe r Cycle (cycles) of F20 ,000 F50 ,000 F100,000 F200 ,000
(ksi) F rac tu re-/c;lnlc;
not included in average; fatigue strengths were calculated only for sound welds
i',/'~i'\
Plain Plate As-Rol led 0.149 89.9 78.5 67.9 60.9 43.6
Transverse Butt Weld As-'We I ded 0.340 77.3 56.3 44.6 35.0
Transverse Butt Weld Reinforcement Removed 0.146 82.0 71.2 64.3 58.2
_'~
,',
k = 0.210
,'"I,
b: failure initiated in weld metal
TABLE .4.2
-;I~/I\
Defective Area, Percent
Spec i men Stress d Radiographic Description
e ,I.
...),
based on radiographic requirements of Navy Specifications(5)
...,',...,1,
effective diameter of largest pore in porosity clusters
TABLE 4.3
-'.
Computed Fatigue Strength," ksi
Spec i men Stress Life Location
Numbe r Cycles (cycles) of , , .......... 1,.,
... ... F20 ,000 F50~000 FIOO,OOO
(ksi) Fracture
...,',
k = 0.210
it,,':.:
b: failure initiated in weld metal
TABLE 4,4
../~
based on radiographic requirements of Navy Specifications(5)
TA BLE 4.5
,'\
data from Ref. 4
i'\'i~\
assuming K = 0.24
(I) subjected to bending
TABLE 5.1
Z-IOO o to +50.0 2,500 5,200 17 ~ 300 ,481 2 - .05 "ea. drilled holes
(:45)(d) (lack of fusion)
Z-101 o to +50.0 90,000 99,400 104,400 .905 2 - .05 ea. d rill ed ho 1es
Z-102 o to +50.0 82,100 89,000 111,500 .922 II - .05 ea. dril led holes
Z-70(e) o to ~50. 0 43,700 52,300 55,600 .836 .14 slag
Z-79~e) o to +50.0 17~000 26?300 32,900 .646 .16 slag
Z-80(e) o to +50.0 11,400 24,700 27,200 .462 .31 slag
(a)
estimated as 1/2 no. cycles between last x-ray showing no fatigue crack and first x-ray showing crack
(b)
estimated as 1/2 no. cycles between last observation showing no crack on specimen surface and first
observation showing surface crack
(c)
defined in Chapter I (usually less than complete fracture)
(d)
fatigue crack initiated ~t lack of fusion, not at"drilled holes
(e)
data from earlier report(4)
-51-
TA BLE 6.1
TABLE 6.2
TABLE 6.3
TABLE 6.4
../~
Thickness of HY-80 Steel Plate
Samp 1e
1-1/2 in, 3/4 in.
0.02
2 0.02
3 0,03
4 0.02 0.04
TABLE 6.5
."
Thickness of HY-80 Steel Plate
S amp] e
1-1/2 in, 3/4 in.
3A 13
crack a 0.08
crack b 0.11
3Al 0.20
4D5 0.26
2D10 0.50
5"
1
I
----r-
-------- 4"
3" __-----:---.i.
I"
I T6 Dio. Holes
t
r' d~1\7"Rod.
I i \
ED CD e e ! I Gl e (9 ~
G (1) -+(01+---
e e e e
L -_ _ _ _ _ _ _ _ _ _ _ _ ~_ _ _ _ _ _
D See flgs.2.2 8 2.3 for weld detail,;
~~------~------------~I-~
t1"
4
~ 5" ~
11
I \\ 7 Rod.
I -----e---e~-~---e~.
(9 (D' e e
J *
J I ... 5
I I
:3 - I
3'
I
I
!
T 2 I
I
!I
I I ... 4
. 6 I
I
:,
I. 6" Continllous Weld Passes .. I
Root opening 13 in.
Arrows indicate direction of weld ing.
Surface of plate adjacent to weld cleaned by grinding before welding.
Pass Electrode size, in. Current, amps. Rote of travel, in. / min.
I i 130 5
32 ~
2 c 140
~ '- 5
3 2- 230 8
16
4 l 220 7
16
5 l 210 7
16
6 l 210 1
16
Root opening t In
Arrows Indicate direction of welding
I i 130 5
32
..5
2 32 140 5
-
3 l 230 8
16
4 .l 220 7
16
_.- .-
5 3 210 7
IS
--~-~
6 ~ 210 7 I
16
I
VOl tage . 21 Volts
Polarity. o C. Reversed
0
Preheat 150 F
Electrode MIL 11018
0
I nterpass Temperature 200 F (Maximum)
Heat Input 40,000 Joules/ In (Maximum)
All welding in flat position
Underside of pass I ground befGfe pioc,ng pass 2
,I I
17,19,21,23
11,13,15
J I
I
7,9
1 ~
5
I
2 I
I
J
J I
I I
3
4 I I
I I
6 I
8,10 I I
I I
12,14,16
18,20,22,24 I
I
I
I
5
1 32
140 4
3
2 16 230 6
5
3 32 160 4
3
4-16 16 230 6
3
17-24 16 200 5
Vol toge: 21 Volts
Polarity: D. C. Reversed
o
Preheat: 200 F
Electrode: MIL 11018
interposs Temperature: 200F (Maximum)
Dynamometer
Specimen
--
80 ~-
N~
-~ ~.... .. ,
.
1;::--
60 ~ .-
~
~- -- - --- --
~
'-'
.. -.J. V
/
r- r-k=O.20
.. r0- t--
~~
--
1- .... ~
40
(J) k = 0.30-./
.........
r--- -- I"'"- __ ..
'-
.lC:
(J)
(J)
-
Q)
....
( J)
E 20
::::J
E
)(
0
~
10
_. Stress Cycle: Zero-to- Tension
I
8 - 0
3"
HY-IOO
4
- 3"
----- "4 HY- 80
6 r-
,
FIG. 3.3 RESULTS OF FATIGUE TESTS OF AS - ROLLED HY-IOO PLAIN PLATE SPECIMENS.
NOTE: FRACTURE DID NOT START AT CORNER.
80 r - - - - - f--.
-
-.-~---
"
I
--1--
....... I
V-- k = 0.297
60 --r-- -_.
~
"' ........
......
.... r--. - I
W ~ 1-,0,.
~
r---- ......... ........ /
................
JI
~ ~ .........
~ r-....
I
.........
(f)
40
k = 0.394 r-- V "" ~ ""'" .........
r--....... ....
~
..::c. ~ .......
~
(f)
(f)
-G,)
b.
( J)
E 20
:J
E
)(
0
~
Cycles
,
To Failure, In Thousands
FIG. 3.5 RESULTS OF FATIGUE TESTS OF HY-IOO TRANSVERSE BUTT WELDS IN THE
AS-WELDED CONDITION.
a) FAILURE INITIATED AT EDGE OF WELD REINFOR CEMENT.
1=U~~~~~IT~
I _
r-
80 -
o~
-
.c
t J')
c:
-
Q)
'-
V>
GJ
60
I
\- .L~~
'0
'r\'{-~...--
2. __i
,..-'--"'-
_
-
::J ~
-
tJ')
o
LL
-~~
....-
~"'F"
--,""
I
-
c: 40 1___+--t-.::--.11--:-:-\-\-::;'(-=_
I L
\\(O~ ~ ...... --~
1"'-
-..... _r _
c: ~ \n __ ....... 1 ~.... ~
-
o
( .)
- - I-- .-, , ___ /
.....- -,- _>--r-
~ 1~~~_~:~~=4~-+-tt1----t-~----r-~--
::J '-
-0
~ \n. Ll'i~O
Q)
Q:
~~. / ,!.-.,....... _ _
~
20
L_ _
... f'" --- I I
'FIG. :3.7 REDUCTION IN FATIGUE STRENGTH FROM AS - ROLLED PLAIN PLATES DUE TO A
TRANSVERSE BUTT WELD IN THE AS-WELDED CONDITION.
100r.------~----~--,_--r_Ir~_,~------_r----,_------._--_,----r__.--.__._._.------II----u_------r----.---,---,
80 1 I ~ I 1- i':: I I I I I I ~~
60
4--I-k = 0.21
40
(I)
..lI::
~1-4-
(I)
(I)
-....
Cl)
en
E 20
:::J
E
)(
0
~
101 I I
t Stress Cycle
8[ Zero - to - Tension
6+-
See Fig. 2.lc for specimen details
5'
2
I 4I 6
I I I I I~
8
I
20 4
~
6
~ I
80
I I
100
I
200
I
400
g
600
FIG. 4.1 RESULTS OF FATIGUE TESTS OF 1-1/2 IN. TRANSVERSE BUTT WELDS CONTAINING
VARIOUS AMOUNTS OF INTENTIONJ~L POROSITY (REINFORCEMENT REMOVED).
a) POROSITY IN WELD.
b) SLAG IN WELD.
-c:
Q)
E
Q)
0
to-
O 90
-...
'0::wc:
-
..c.
'i
en 80
3.2 I
Q)
~
,
"'0
c
~
0
(f)
70 (Butt Welds - As Welded)
-... 0 0
0 0
0
0
0 .. u u
8 )
o 0
lJ...
60
\6-
-0
c
Q)
0
"- E
"'0
Q)
>
0
r-
Stress Cycle
Zero - to - Tension
Note: Diameter is based on maximum
0
OJ OJ
a.. a:: pore on fracture surface.
--~~ I I I I
50
o 0.01 0.02 0.03 0.04 0.05 ~ 0.06 0.07 0.08
FIG. 4.3 PERCENT FATIGUE STRENGTH AT 100,000 CYCLES VS. EFFECTIVE DEFECT
DIAMETEH OF POROSITY IN 1-1/2 IN. MATERIAL.
100
CI) 80
~
~
CI)
CD
~
U (Butt Welds - Reinforcement Removed)
o
o 60
o.
o
o
....o 00
(Butt Welds - As . Welded)
0
:ED 40 1)~ <) 0 .....,
C
CD
C
....
t...
(f)
QI
:::s
-
~
0-.
20 ------ ~- 1 - - -r--'
Stress Cycle
Zero - to - Tension
Note: Area is based on the defect
area shown on radiograph.
o
.02 .04 .06 . .08 10 20 .40 .60 .80 1.0 2.0 4.0 6.0
FIG. 4.4 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES vs. DEFECTIVE AREA DUE TO
INTENTIONAL. PCROSITY IN 1-1/2. IN. MATERIAL.
100r-=r--------r----~----~--~~--,_~~--------~-----~------~r_----r_--_r--_r--,_~--r_~-------~----~:-------~----_r----r__,
80
40+--
en
.lII::
U)
U)
-I"+-
-~
00
E 20t--
I I I I
::J
E
)(
o
~
,0~----~--1---L-JL-L-1-L4------~--_+----_4--~~_t--t_i_i_t_r_----t_~1l----_t--~--1I~
t- Stres s Cycle:
8-1- Zero - to - Tension
I-
6+-
See Fig. 2.lc for specimen details
.i_~_ I __ I J _ I I
52 . 4" 6' S" . 10 20 40 60 80 100 200 400 600
FIG. 4.5 RESULTS OF FATIGUE TESTS OF 1-1/2 IN. TRANSVERSE BUTT WELDS CONTAINING
VARIOUS AMOUNTS OF INTENTIONAL SLAG (REINFORCEMENT REMOVED).
/00 -
CI) 80
.lI:
CI)
.
CD
~
U (Butt Welds - Reinforcement Removed)
oo 60
o flO
o
o lA
IU
....o b.
(Butt Welds - As' Welded)
fA
.c:. b. b.
~ 40 L
l.
c
Q) 0 ~
....'-
en
OJ
:::J
-
.~
~ 20
Stress Cycle
Zero - to - Tension
Note: Area is based on the defect I
_~grea show~~__ !adiog~aph.
o - - '----~~~ -~ '---------------~ '-------- I
.02 .04 .06 .08 ./0 .20 .40 .60 .80 1.0 2.0 4.0 6.0
FIG. 4.6 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES vs. DEFECTIVE AREA DUE TO
INTENTIONAL SLAG INCLUSIONS IN 1-1/2 IN. MATERIAL.
100 .,
~ "-
U
bt. J.P.
o
o 60 Do
~
Do
o f.
bt.
__
ci
N 0 ~
-C
-
.t::.
en
c
40
-Q)
'-
UJ
W
::J
-
.2'
0
LL
20
Stress Cycle
Zero - to - Tension
Note: Area is based on the defect
area show~ on ~CJdi9grc:Je~~___ ~
o -~-
.02 .04 .06 .08 .10 ..20 .40 .60 .80 1.0 2.0 4.0 6.0
FIG. 4.7 COMPUTED FATIGUE STRENGTH AT '20,000 CYCLES vs. DEFECTIVE AREA DUE TO
INTENTIONAL SLAG INCLUSIONS IN 1-1/2 IN. MATERIAL.
ELEVEN DRILLED HOLES IN A ROW.
en 80
~
en
.. ,-
CD
~
U
oo 60
o
ci (Butt Weld - As . Welded)
lO
....o
.c.
1; 40 U
IE:: 0 /::,. L ~
Q)
/::,.
....
'-
(J)
0
GJ
::J
A
-
.~
~ 20
o Porosity /::,. Slag
Stress Cycle
Complete Reversal
Note: Area is based on the defect
area shown on radiograph.
o
.02 .04 .06 .08 .10 .20 .40 .60 .80 1.0 2.0 4.0 6.0
FIG. 4.9 COMPUTED FATIGUE STRENGTH AT 50,000 CYCLES VS. DEFECTIVE AREA DUE TO
INTENTIONAL WELD DEFECTS IN 3/4 IN. MATERIAL.
100
en 80
~
en
..
Q)
~
U (Butt Weld - Reinforcement Removed)
o
o 60
o
o
o ~
...
o
~
00 0 ~ fA 1 - r--'"
(Butt Weld - As' Welded)
.s:::. ~ ( o _~
en 40
,.....
~
fj L:.
c C
~
...
en
0
cu
:s
-
.~
If 20
o Porosity ~ Slag
Stress Cycle
Zero - to - Tension
Note: Area is based on the defect
area shown on radiograph.
o
.02 .04 .06 .08 .10 .20 .40 .60 .80 1.0 2.0 4.0 6.0
FIG. 4.10 COMPUTED FATIGUE STRENGTH AT 100,000 CYCLES \IS. DEFECTIVE AREA DUE TO
INTENTIONAL WELD DEFECTS IN 1-1/2 IN. MATERIAL.
a) EQUIPMENT AND LEAD SHIELD IN POSITION.
en
(1)
a.
.\
0
~
Slag
0
Q)
Q.
c: "
I
Porosity
Drilled Holes
(J)
~
0
0 I
\\ Lack of Fusion .JIC:
0
0
'- 80 '-
U
(1)
\ u
W
::I ::I
.Q'
+- \ (0 to + 50.0 ksi) .~
0 +-
0
LL
~
0 60
\ LL
~
0
0
c:
\ c
.2
.
+- +-
.2 .Q
\
+-
:!:
c (I) c
~
\
.EO
"0
40
" 0
10 -
0
(/)
".
(/)
~LO
" "
0 "O~
en + 0
.
::::lI
0
L:.
0
+-
'\, I
I
(/)
::1
oLO
0
..... 0 I L:.
..... -H
c
- 20 u 5
c
-
"
0
0
~J
en 'U
~/). .............. 'U
(1)
-
QlI (/)
- +- (1)
W
U (/)
to-
----
(/)
>. w (,)
U ..... /). >.. w
u .....
- -I-- .
o
o .10 .20 .30 .40 .50
FIG. 5.4 COMPARISON OF CYCLES OF FATIGUE CRACK PROPAGATION TO TES LIFE IN 3/4 IN.
HY-80 TRANSVERSE BUTT WELDED SPECIMENS TESTED AT + 50.0 KSI.
40
Specimen Symbol
(f)
Q)
Z - 98 /::,. I
I
Z-IOO
E ~ z -101 "
f--------+---t----------I ---------t----
ou 0
'-
lL. ..... Z-102 0 I
o 30 f--------.J
c
o
:;:
CJ)
"0 Z-70 * A
o c
0\ 0
o CJ) Z-79 "* I I I
a. :::J
o 0
'- ..c.
a.. I-
l-80*
-* Data
from prevIous study
..lit: C 20
g I
u'- Q)
~
--
....J
Q)
:J i~
0\
:;:
o
-CJ)
Q)
--- ------
0
0
c
--
0
0
'-
c
0
0
V-- --- --- ----
:::J
0 C
o
o 20 40 60 80 100 120 140 160
FIG. 5.5 COMPARISON OF CYCLES OF FATIGUE CRACK PROPAGATION TO TEST LIFE IN 3/4 IN.
HY-80 TRANSVERSE BUTT WELDED SPECIMENS TESTED AT 0 TO 50.0 KSI. +
__-------------------~----~---------~--------,_--------_r~-~~~m~~~~t~~~~~~~
(J)
w
.c.
u
c
~
"..
(/)
.c.
a.
I I
o I ~
r...
0' en
~2 Q.)
"0 I-
o I 201--------+---------+-
0::
c Region f Fatigue .. ,.:Y<~~~~
o Crack P opagation. .r,l:~P;-~;Pf
- ~:~:,~~,;P)ffip,1~~ftt
"0
W
o
.~ .801 , ..
"0
c:
..:.:: F_ ;>
u Cra
0
r...
U
\4-
0
-
.c:
C"
c:
QJ
.40
-I
Cycles Prior to Appearance of Surface Crack Cycles Beyond Appearance of Surface Crock
FIG. 5.6 RATE OF CRACK PROPAGATION IN 3/4 IN. HY-80 TRANSVERSE BUTT WELDED SPECIMENS
TESTED AT 50.0 KSI.
2.00 .
Spec I me n S ym b 0 I
T .. :'t.>n~ .. ':'\.i:,';<~:."h -; .. ! ......, ,I(L ', . . ~~;......" '~ ,:.~ .., .:.... ,~,
, ..
~~''''~)~;.~ ~P.~.::at'~:~:\(~~~.J:'~(~~~:'rt~~f~W<F.lij~W.~~;(l.;;:".;;~h.:9::1.~~it.:}>:':::.
:"";.,.",:),",', ""';":.~. ~'<. t', 0'
~ l;:i~ ~ J_-~_-_--L---
V)
..
..r::.
1.60
1 .;,IIIIJliltllltir'lili!
Z-70~..
A I
~-1
W
C-
o z-79'*. I!
"-
0"
Z- 8 0 * . I (/)
o
.--1-- . ___
---'- W
'0
o
0:;:
120 *" Dota from prevIous study --------111
..- ~,.... ~ r." ' ' ' _~' ~'.'<.~' ~':
,.! ......... .... r''''''''..... _...! .: . k.'''''!t ... <:.i>J\ ... .,,~. ~."'Jf'.\<.'
... ,.0( ..... _ ... . .... 1'.". 1:1I'
i
c F atiguc
o
...o
"0
Q)
,~
"0 .so ....- - - - i -
C
..:
U
o....
U
~
o
-
..r::.
C\
C
w t-F~0~w\n:f
. II S oa~ - Type II t
...J
o Crack l-'r8Do. otlon .
o:t:~~tlt{ }t~~~~&~tiiif~1?;\~%~'/" ~._______~II-________ ..&..._ _ _ _ _ _ _ _..A-_ _ _ _ _ _ _ _.....J._ _ _ _ _ _ _ _- ' ' ' ' ' '_ _ _...II
Cycles Prior to Appearance of Surface Crock Cycles Beyond Appearance of Surface Crock
FIG. 5.7 RATE OF CRACK PROPAGATION IN 3/4 IN. HY-80 TRANSVERSE BUTT WELDED SPECIMENS
TESTED AT 0 TO 50.0 KSI. +
Section
A AA
Note A
7-2I"
4..L"
2
.',
~
60
~
c::
~ ..
24"
f."
::i:t;
I" ,7 ... )','",".
4- .... -..
L
,'I"t',
~J_'I~:.
l :"':.::~
HY - 80 Steel Plate
FIG. 6.1 WELDING JIG USED TO RESTRAIN HY-80 STEEL DURING BUTT WELDING.
INGOT TOP
t
36" A A-3 A-6
130"
t36" A-I A- 4 A-7
+-L
36" A-2 A-5 Main Rolling Direction
.....
l I~ 144" .1I ~
- - - - - 440"
1 96" .1 j
I
INGOT BOTTOM
FIG. 6.2 MAIN PLATE OF HY-80 STEEL FROM WHICH TEST PLATES G-67, G~68, AND
G-69 WERE OBTAINED.
LO'IGITUD I NAL AX IS OF WELD :>
TOP VIEW
o *"DISCARJ
IN.
o SAMPLE
~IN.
71~------------+-------------4--------------r----------~-+-------------1--------------r-------------4-------------4-------------~
61 Z 87 Samples
I
>.
uc: _
~ 41 _
0"
4 W. U. I Micron
~ 3
ll..
8.r------------~-------------..,._-----------------~----------------~-------------~----------------~----------------r-------------~~------------_....
71 -1-------------+-------------1-------------~------------_;
61 _
G - 69 Samples
>.
u 511-1-----
C
Q)
:::s 41----..,.-
0"
~ 31 _
ll..
2
0
0 fOO 200 300 400 500 600 700 800 900
:>.
u 6i-----
c:
Q)
::J 4 T. U. = I Micron
0-
~ 4~--
lJ..
o
o 5 /0 15 20 25
14
12
G - 69 Samples
10
:>..
u 8
c:
Q.)
::J
0-
W 6
'-
lJ..
0
0 5 10 15 20 25
OF TH WE LO. 400 X e
FIG G.U PHOTOMICROGRAPH OF A CRACK ALONG
500 _L_ 5
...... , ........................... I -
A
I
I 6
I
CI)
CI)
Q)
400
IJ\,,: I
:1
,/
-
l
C
"0
I 0 40
Y
"-
n
0
I
:'Q
a..
E
0
"-
~
300
I,
I I
U
"0
c I I
E
0 ....",.,.,..
~ Normal
I I "Buttered" ~~~
0 I I
Base Metal HY-80 HAl Weld Metal HAl Base Metal HY - 80
0 200 F.I, F.I L.
I I
I I
100 -
a
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38
FIG. 6.12 MICRO HARDNESS SURVEY OF 1- 1/2 IN. THICK HY - 80 STEEL BUTT WELD.
500 T
I I I I I
I
I
I
!(!'J\
..........
I
I
~
40C F.IL. -
~
I
(f) I
~
(f)
Q)
I
~)
C
"0 II Buttered II
~
o
1~F\ HAZ
I
"0
30C
VV VI
A
I
E I
o
~
~
I
I
k --........... ~ V
.........
~ .........
-
0.. I
"0 Weld Metal I Base Metal HY -80 Steel
C 200
o I
E F.IL.
o
o I
I
10C
01
o 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
FIG. 6.13 MICRO HARDNESS SURVEY OF THE WELD METAL AND HAZ OF "BUTTERED"
HY-80 STEEL SCARF.
FIG 6.14 FRACTURED SECTION OF HY'" 80 STEEL
IN FIG 6.14.
FIG 6.16 PHOTOMICROGRAPH SHOWING FATIGUE CRACKS
APPEND IX A
Ultrasonic Crack Detection Stu~y
are presented in Table A. 1, and plotted in Figs. A.l and A.2 for tests
report, ultrasonic readings were taken fol lowing each X-ray at various intervals
during the 1 ife of each specimen. The ampl itude of the response as shown on
the oscil 1 iscope screen was recorded for each of eight consecutive locations on
location nearest the point of actual crack initiation (as shown by the radio-
locations away from the point of initiation showed considerably less fluctuation
unti 1 the crack had progressed for some time. It should also be noted that
Superimposed on each plot in Figs. A. 1 and A.2 are the points where
the radiographs first showed the existence of a fatigue crack. The results
indicate that the ultrasonic system of detection does show promise as a future
Cycles Ultrasonic Cycles Ultrasonic Cycles Ultrasonic Cycles UI trasonic Cycles Ultrasonic
read i ng reading read i ng read i ng reading
32 16 26 0 34
2,000 29
1,000 20
0
2,000-
30
35
0
2,300- 28 2,000- 31
97,000 avg. 69,000 avg. 67,200 avg.
4,000 25 2,000 26
103,000 33 76,000 31 77, 100 32
6,000 39 3,600 24
109,000 36 85,000 35 87,000 29 I
8,000 37 6,100 32
N
119 ~ 000 50 95,100 50 91,000 58
10,000 40 7,600 36
121 ,000 50 99,500 54 93,100 58
11,600 56 10,000 36
102,000 57 95,000 50
12,200 53 11,600 41
104,400 62 97,000 56
12,800 53 13,100 41
99,000 49
13,800 48 13,200 46
Specimen Z-96 45
40
35
Ol 30
c:
"'0
o 25
(l)
a::
(l)
a. 20
o
u
(f)
60~------~------~------~------+-------+-------+-----~15
-
~
o
u
-
(l)
( lJ
o
55
50
Specimen Z - 94
u
c:
o
(J)
45
o
-~
40
35
First indication of
fatigue crack as
shown on radiograph.
25
20~------~------~------~------~----~~------~----~
o 2 4 6 8 10 12 14
Cycles, In Thousands
0\ 40
c: Avg. Magnitu~
-e--_ .......
"0
o 301 160
Q)
~
0::: Specimen Z -101
Q)
a. 50
o
u
en 40
--
'-
o
u
Q)
Q)
:30
o 651 120
.S? Specimen Z - 102
c
o(J)
55
-o
'-
:J 45
\. First indication of
fatigue crack as
shown on
radiograph.
35
Cycle s, In Thousands